S.

Weber

377

Curing of high strength concrete using lightweight aggregates

SILVIA WEBER
Bundesverband der Deutschen Zementindustrie e.V.
Bauberatung Zement Stuttgart, Leonberg

Abstract
A technical solution for the still existing problem of curing high strength concrete is proposed. The
new method consists in using a blend of aggregates containing lightweight aggregates. The
prewetted grains become a water supply at the disposal of the hardening and drying cement paste,
allowing a continuous hydration which leads to improved properties of the concrete.

Introduction
An essential way to achieve the designed properties of concrete is avoiding water
evaporation at the surface and supplying water from the exterior. If enough water
is at the disposal of the cement paste for hydration to proceed, the concrete will
achieve excellent properties. The traditional ways of curing often failed in practice. Even when meticulously performed only water evaporation can be reduced,
but the water supply on the surface of vertical structural elements is still a technical problem. The time allocated for curing is a stagnation of building time increasing costs and efforts. The efficiency of modern technology like climb and
slip forming is perturbed and the risk of damages caused by improper curing is
still not eliminated. In case of high strength concrete (HSC) used commonly for
vertical structural elements the problem increases. Due to a very low water cement
ratio in combination with a high cement content and the addition of silica fume
the concrete shows a high compressive strength at early age, which makes an
early form stripping possible.The high self-desiccation was the reason for using
this concrete for self-desiccating slabs [1]. The very dense structure of high performance concrete might lead to the assumption that water evaporation is low and

378

S. Weber

therefore water from the surrounding can penetrate only very slowly and not in a
sufficient amount to achieve the interior of the member. So, curing can therefore
be neglected. Saving the curing time would influence positively the construction
costs. On the other hand when exposed to air water evaporation was observed [2]
resulting in a considerable reduction of the compressive strength [3] and microcracks appear [4]. Own researches [5] showed for HSC a failure of the traditional curing methods. Wet curing for a longer time- even submersing in waterdid not always influence positively the mechanical properties. Hence, the opinions
and literature data about curing high strength are contradictory.

Theoretical considerations
The main idea was to create in the concrete a water supply which is independent
of the environment and which offers to the unhydrated cement water to continue
hydration processes and to equalise the humidity content in cement paste during
drying and hardening. Lightweight aggregates (LWA) generally can store a high
amount of water. Their uniform distribution in the cement paste and a continuous
transport of the water from the LWA to the cement paste transforms each grain of
LWA to a water supply for the concrete surrounding it. The water transport in
concrete is depending on the microstructure of the concrete and on the existent
humidity gradient. As soon as hardening of concrete occurs, in the cement paste a
system of pores develops. Silica fume concrete shows a refined pore structure
with smaller radii than in normal concrete. The radii of these pores are smaller
than the radii of the pores in the LWA, so that the requirement for water transport
is the existence of a humidity gradient. During hardening and/or drying of concrete humidity gradients are distinguished on microscopical and macroscopical
level. On microscopical level a shortage of water occurs due to chemical reactions. The capillary forces of the cement paste are high enough to absorb the water
from the LWA grain and transport it to the dryer cement paste where a reaction
with the unhydrated cement takes place, forming Ca (OH)2. After the reaction of
Ca (OH)2 with silica fume new hydration products of type calciumsilicate hydrate
(C-S-H) are built, growing into the capillary pores, the available space, and into
the micrcrocracks as well, making the structure of the cement paste more dense.
As the C-S-H have a higher strength than Ca (OH)2 has, the compressive strength
increases. As the suction forces in the capillary pores are inversely proportional to

S. Weber

379

the radius, the smaller the capillary pores of the cement paste get, the higher the
suction forces in the smaller pores. Thus the difference in pressure (as the condition for water transport from the LWA to the cement paste) is given. This transport will stop when the relative humidity in the LWA and in the hardened cement
paste are in equilibrium. This is a time depending transport of at the begin water
and in the end vapour, with higher rate in the early age of concrete and lower at
later age. On the macroscopical level the humidity gradient on the surface of a
concrete member exposed to the environment is considered. The lower the relative
humidity of the environment is, the steeper is the gradient between the surface
layer and the more distant layer. Due to water evaporation the gradient increases
and in the surface layer the moisture from the LWA will be transported to the cement paste faster than in the interior of the considered member. Because the water
from the LWA is chemically bound, the structure on the surface is more dense,
reducing the water evaporation and the diffusion process becomes slower. The
surface of high strength silica fume concrete with a blend of aggregates being
more dense than the traditional HSC, the exposure to dry environment at early age
will not have a negative influence on its properties.

The task of the research was to transfer this idea into practice, developing a high
strength concrete containing partly LWA which can be form stripped at early age
and achieves improved properties without applying any curing.

Research program
Material Properties and Optimisation Procedure
The optimisation was performed starting from a reference mix of high strength
concrete containing : 450 kg/mcement type CEM I 42.5 according to [6] , 30 kg
dry mass of silica fume added as a slurry, 13,6 l/m superplasticizer (sulphonated
naphthalene-formaldehyde condensate) and 1,75 l/m retarder. The normal weight
aggregates (NWA) in amount of 1735 kg/m are dry rounded sand and gravel
from the upper Rhine valley with a maximum size of 16 mm, separated in fraction
as follows: 399kg/m fraction 0/2, 208kg/m fraction 2/4, 399kg/m fraction 4/8
and 729 kg fraction 8/16. The water/cement ratio was 0,33. The fresh density of

380

S. Weber

this concrete was 2350 kg/m and the workability measured on the flow table ten
minutes after mixing a10 >50 cm. The compressive strength c,100,28 =104 MPa was
determined on 100 mm cubes at the age of concrete of 28 days. The cubes were
cured according to DIN 1048 [7] which means 6 days submersed in water and 21
days at air by T= 201C and 65% RH.
For the optimisation the cement and water content were kept constant while the
normal weight aggregates were partly substituted by several types of expanded
clay LWA, each replacing completely a certain fraction. As shown in Table 1 and
Table 2 the LWA differed in grain size, strength, density, porosity and water absorption properties and hence moisture content.

Table 1: Properties of lightweight aggregates by sprinkling

Code

Grain Size

Dry Density

Moisture [% by mass] after hour of

[mm]

[kg/m]

Submersing*

0/2

1516

19,8

2/4

950

8,2

7,9

3/6

1300

11,1

16,3

4/8

930

11,8

12,5

4/8

1420

6,1

7,8

sprinkling

*declared by the producer

Table 2: Properties of lightweight aggregates wetted by submersing

Code

Grain Size

Dry Density

Moisture [% by mass] after

[mm]

[kg/m]

2/4

950

8,2

20,8

3/6

1300

11,6

4/8

930

11,1

18,3

4/8

930

11,1

16,5

4/8

1420

6,1

20,2

hour

24 hours

There are two main possibilities to introduce the LWA to the concrete mix: dry or
wet. When introducing dry grains the water absorption of the LWA has to be
known in advance and taken into consideration when calculating the water to be

381

S. Weber

added. A disadvantage is, that the water absorption during mixing and hardening
cannot be exactly determined. Additionally, the moisture absorption of LWA in
fresh concrete differs considerably from the water absorption [8]. Considering that
the water added suplimentarily to the mix is not absorbed by the LWA and can
evaporate, the intended water reservoir cannot be achieved by this method.

Wetting can be done by submersing or by sprinkling for a given time. In this cases
the moisture content can be determined in a simply way by drying, but one has no
indication of the amount of water adhering on the surface of the LWA, which has
to be considered by calculating the amount of water to be added. The water from
the surface can be determined by the CM-Method commonly used to determine
the moisture content of sand. As sprinkling the LWA for half an hour led to high
scattering in the moisture content submersing for 24 hours was performed.

With the LWA shown in Table 1 and 2 different concrete mixes were designed.
Table 3 shows the type and the amount of the used normal weight aggregate and
of LWA in % of the total volume of aggregates. 1 m concrete contains 450 kg
cement, 150 l water and 1,75 l retarder.

Table 3: Mix proportions per m concrete

Aggregates
Mix

NWA

Lightweight

[%] by vol.

Super-

Silica

plasticizer

Fume*

Type

Replaced Fraction

[l]

[kg]

75

25

sand

6,8

30

II

90

10

2/4

6,8

30

III

80

20

2/4

6,8

30

IV

90

10

3/6

6,8

30

80

20

3/6

6,8

30

VI

85

15

4/8

13,6

30

VII

85

15

4/8

13,6

30

VIII

85

15

4/8

13,6

30

IX

85

15

4/8

13,6

30

80

20

4/8

13,6

45

dry mass of the slurry

382

S. Weber

The preliminary tests performed on 100 mm cubes cured in the folowing conditions: KK in air at temperatures T between 15C and 25C, relative humidity (RH)
varying from 40% to 45%, KR in air at T=20C, RH=65%, KL sealed in aluminum and polyamid foils and NK submersed in water for 6 days then in air T=20
C, RH=65% showed for all mixes high compressive strengths.

Mix VII of the showed the best workability and a uniform distribution of the
LWA in the cement paste. The LWA of type J has the advantage of a 115 MPa
strength combined with a porosity of 50% and a water absorption of 20% of the
dry mass when submersing. Figure 1 shows the pore size distribution [9].

35
30

dV/d(logr)

25
20
15
10
5
0
0,001

0,01

0,1

10

100

Radius of pores [m ]

Fig. 1: Pore size distribution of lightweight aggregates type J [9]

As by the reference HSC a relative high water evaporation was observed which
should be covered as well by the water supply, the required amount of water to be
stored in the LWA was of 47 l. It was necesary to replace completly the fraction
4/8 mm, which corresponds to 234 kg. The remaining aggregates are normal
weight aggregates in amount of 1298 kg/m separated in 471 kg/m of fraction
0/2, 260kg/m of fraction 2/4 and 567 kg of fraction 8/16. The amount of silica
fume was increased to 45 kg dry mass /m. The dry density of the concrete is 2150
kg/m, the workability a10>50 cm and c,100,28 = 104 MPa.

383

S. Weber

Modelling the water transport in the cement paste

Water Transport within the Cement Paste
The results of the mercury intrusion are given as a differential distribution of the
pore volume versus pore radius. Logarithmic scales are used. By plotting dV/d log
r versus log r the area below the distribution curves correspond to pore volume.
Using these curves the developed pore system of the cement paste can be defined
by the equivalent pore radius [10] like shown in Equation (1):

rmax
rmax

dV
dV
2
2
r eq =
d log r
r d log r /

r d log r
r d log r

min
min

(1)

The equation of the equivalent pore radius is relatively correct for concrete with
pores of smaller radius [10]. Using the transport process in a single capillary pore
of radius req the water transport of the pore system can be simplified modelled.
Assuming a variation of the radii in the pore system with the tendency towards a
greater number of pores with smaller radii, req is time depending req = f(t), being
smaller with increasing age of concrete.When hardening of concrete starts the
equivalent pore radius existing at early age of concrete can be considered the
greatest equivalent pore radius req,max while at the age of concrete when all the
water to be transported is used up, the smallest equivalent pore radius req,min dominates. In the meantime the number of pores is considered to be constant. At the
beginning of hydration all capillary pores described by req,max are filled with water
and due to the capillary pressure and the tension developed at the surface of the
meniscus in the fluid a hydrostatic equilibrium is established. When hydration
and/or desiccation occurs the pores become smaller achieving at a certain moment
t a pore radius req,t < req,max, and a difference in capillary pressure appears. At any
time t the capillary pressure in the pores of radius req,t is given by:

PK

(t )

2 cos
req( t )

2)

384

S. Weber

Water transport from lightweight aggregate to the

cement paste
For the concrete with lightweight aggregate an equivalent pore radius for the pore
system of the LWA pores R

eq,LWA

can be described like in the equation (3) by

adapting Eq.(1) and similar the capillary pressure PLWA in the pores R

eq,LWA

by

Eq.(4). The number and size of the pores are constant in time.

R max,LWA
R max,LWA
dV
dV

R2eq, LWA=
dlogR
R2 d logR/
d
log
R
d
log
R

R min,LWA
R min,LWA

PLWA =

2 cos
R eq ,LWA

(3)

(4)

To show the moisture transport from the LWA to the cement paste a model of two
cylindrical pores in contact is used. The cement paste can be represented by a
capillary pore of radius r

eq(t)

and the existing water supply of the considered

lightweight aggregate by the equivalent pore radius Req,LWA . Figure 2 shows the
different stages of water transport described in the following.

Fig. 2: Modelling the water transport from the LWA to the cement paste

When hydration proceeds, the smaller pore exerts suction forces in the greater
pore Req,LWA producing there a lower level of fluid, located at the distance L1. The
more the meniscus of the pore with radius Req,LWA retreats, the higher is the capillary tension in the smaller pore. When this tension exceeds the friction forces in

385

S. Weber

the smaller pores req(t), the meniscus change its position. The difference of the
capillary pressures P= PK(t) - PLWA makes the capillary transport on the distance
L1 possible: a further transport of water is now possible only when the meniscus
of the smaller pores shifts into position l1. The capillary transport for this distance:

1
1
m (L 2 l1 ) =

4 req ( t ) R eq ,LWA

2
req ( t ) w

(5)

When the meniscus of the greater pore is in position L, there is no pressure difference P and the remained fluid in the smaller pore is used up either by capillary
transport between pores of the cement paste (which can be described similarly) or
by diffusion combined with condensation. Between the beginning until the end of
hardening this difference is given by the Equation (6).

P =

2
1
1

r
R
eq , LWA
eq ,min

2
1
1

r
R
eq , LWA
eq ,max

(6)

To make a simple calculation possible the assumption that the pores of the LWA
and of the cement paste have the same length l which is known and the fluid water
has a laminar flow through pores is necessary. As the length of the pores are not
known with the aid of another modelling an idea of the distances on which the
water has to be transported can be given, starting by the uniform distribution of
the considered LWA of fraction 4 to 8 mm through the cement paste. In this modelling the volume of 1m concrete can be subdivided into a number n of identical
spherical composite elements of radius RCE consisting of an aggregate of radius ra
surrounded by a layer of cement paste having a constant thickness. In Figure 3 the
modelling of the concrete using composite elements is given.
NWA
wet LWA

cement paste

&HPHQWSDVWHDURXQG1:$
1:$U PP
D DPLQ
/:$5/:$ PP
D/:$
'PD[  D/:$

Fig. 3: Modelling of concrete and distances between the aggregates grains

&HPHQWSDVWHDURXQG/:$

386

S. Weber

For the total amount of the aggregate Z, the radius RCE calculated from the following Equation (7) is given by the Equation (8):

1000 4 3
= R CE
n
3

(7)

10 ra
RCE = 1
Z
3

(8)

In reality the total amount of aggregate contains grains of different fraction. Having a certain number i of fraction sizes of volume Bi according to Eq. (8) the radius Ri depending on Bi can be obtained. The difference of the radii Ri - ri gives
the thickness of the layer ai which surrounds each grain of different size. The distance between two aggregates of the same size is two times the thickness of the
layer. The calculation of the thickness of the surface layer and distances of grains
of the same size are given in the following.

Table 4: Calculation of the thickness layer and distances between grains

Radius [mm] Vol. Bi of grains [l]

10/ (Bi) 1/3

Ri [mm]

ai [mm] di [mm]

r1= 0.0625

6.6

5.33

0.33

0.27

0.54

r2= 0.125

26.4

3.36

0.42

0.29

0.59

r3= 0.25

46.2

2.79

0.70

0.45

0.89

r4= 0.5

6.6

5.33

2.67

2.17

4.33

r5= 1.0

66,0

2.47

2.47

1.47

2.95

r6= 2.0

92.4

2.21

4.42

2.42

4.85

r7= 4.0

59.2

2.57

10.26

6.26

12.53

r8= 4 LWA

165.0

1.82

7.29

3.29

6.59

r9= 8.0

191.4

1.73

13.88

5.88

11,76

On the distance between two LWA grains other aggregate grain can be placed.
Considering that every LWA grain is a water supply for the volume of cement
paste surrounding it, the distance on which the water has to be transported can
simplified be estimated by considering the limit states of maximum and minim
transport distance. The longest distance Dmax would be the most inconvenient case
of transporting the water on the distance between two lightweight aggregates

387

S. Weber

Dmax= a8. This would be the longest capillary pore, when assuming their linearity.
The most detrimental situation for the concrete would be when a normalweight
aggregate is positioned as near as possible to the LWA, being a barrier on the
waters way through the cement paste. As the shortest way in this case is the thickness of the layer surrounding the smallest grain, this situation occurs when a aggregate of r1 =0.0625 m is in the vicinity of a LWA grain. The thickness of the
layer will determine the shortest length of a capillary pore being identical to the
minimum length of Dmin= a1.

Experimental program
On the concrete mix with lightweight aggregates macroscopical and microscopical investigations show the influence of autogenous curing and validity of the
above shown modelling. On 100 mm cubes cured in different curing conditions
the mass loss versus time was monitored and the compressive strength was determined. In Figure 4 the recorded mass change and the ratio of determined compressive strength to the characteristic compressive strength are shown.
120

60

110

50

Ratio /c,100,28

40

..F
.5F

90

30

./F

80

1.F

70

0DVVFKDQJH.5

20

0DVVFKDQJH..

10

0DVVFKDQJH./

60

0DVVFKDQJH1.

50

-10

40

-20

30

-30

20

-40

10

-50

Mass change [g/1000cm]

100

-60
1

10

100

1000

Age of concrete [days]

Fig. 4: Mass change and compressive strength of HSC with LWA

From fragments of the crushed cubes stored in the most unfavourable curing condition KK the physically bound water was eliminated by storage in an exsiccator.

388

S. Weber

Then they were crushed again and particles were picked out which contained only
cement paste and sand, serving to determine the pore size distribution with mercury porosimetry shown in Figure 5.

40
Age of Concrete =360 days

35

Age of Concrete =180 days

dV/d(logr)

30
25
20
15
10
5
0
1

100

10000

1000000

Radius of Pores [nm]

Fig. 5: Pore size distribution of concrete

The other part of the particle was hand milled and sieved through a sieve with a
mesh size < 0.063 mm, obtaining samples rich in hydrated cement paste. The
amount of chemically bound water and thus the degree of hydration was calculated on these samples like shown in Table 5.

Table 5: Calculation of the degree of hydration

Amount of water considered

Age of concrete [days]

180

360

Initial in concrete at t=0

150

150

Chemically bound when =1

121

121

Total chemical bound determined by TG and DTA

82

100

Calculated degree of hydration

0.67

0.82

The results of the simultaneously performed thermal gravimetry (TG) and differential thermal analysis (DTA) are shown in Fig. 6.

S. Weber

389

Fig. 6: Results from TG and DTA, age of concrete 360 days

On samples rich in hydrated cement paste X-ray diffraction was carried out. The
obtained chemical composition at age of concrete 180 and 360 days are shown in
Figure 7, for the reflecting angles of 9 to 23.

Fig. 7: X-ray diffraction, angles 9 to 23, age of concrete 180 and 360 days.

390

S. Weber

Discussion
Regardless of the curing, a continuous increase in compressive strength with later
age of concrete is recorded. The reason of increase in compressive strength independently of the environment lays in the more dense structure of the cement paste
indicated by the porosity measurements, clearly showing a decrease of the total
porosity with increasing age of concrete. Comparing the pore size distribution of
the concrete at age 360 days with the age of 180 days a decrease of the number of
pores with a great radius and an increase of the pores with smaller radius is observed (Fig.5). This refining of the pore system explains the decreasing porosity
and leads to the presumption that water for hydration reaction was indeed available and chemically bound. The higher degree of hydration (Table 5)at later age
of concrete confirms the chemical reaction of the supplementary water with unhydrated cement grains and by X-ray diffraction a higher amount of chemical compounds like calciumhydroxide and ettringite at the age of 360 days are detected
(Fig.7). As Ca(OH)2 results from the hydration of cement water had to be on the
disposal of the cement grain. The X-ray diffraction shows also more C-S-H of the
type afwillite at the age of 360 days, compared to the age of 180 days. The supplementary built C-S-H as well as the higher amount of Ca(OH)2 and ettringite are
the result of supplementary hydration processes. These stays for the continuation
of hydration processes from the age of 180 days to the age of 360 days. These
aspects are confirming the theoretical assumption of ongoing hydration processes
resulting in the predicted more dense structure at later age, due to the partial filling of the pores by additionally formed hydration products. The peak shifting in
the curve of the pore size distribution into the direction of smaller pores stands for
creating higher capillary pressure necessary for the water transport from LWA to
the cement paste. From the mass loss of the cubes versus time a humidity gradient
is shown. From the pore size distribution (Fig.5) and the mass loss (Fig.4) the
transport conditions (increasing capillary pressure and existence of a humidity
gradient) for the shown modelling are given. Compared to the conventional HSC
containing only NWA the amount of chemically bound water is about 24 l per m
concrete higher [5] and represents only the half amount of the water stored initially by submersing in the selected lightweight aggregate. Hence this concrete has
still a water supply at its disposal.

S. Weber

391

Acknowledgement
The results are part of a research program which was carried out in the framework
of AiF (Co-operative Industrial Research) Project AiF -Nr.9816 and co-ordinated
by DBV (German Concrete Association) Project Nr.178. The support by Liapor
Franken, Elkem, and Heidelberger Zement AG is gratefully acknowledged.

References
[1] Persson, B.S.M. Self-desiccating high-strength concrete slabs, In Proceedings Utilisation of High Strength
Concrete, Symposium in Lillehammer, Norway, June 20-23, 1993, eds. I. Holland and E. Sellevold, pp 882889.
[2] Hammer, A.: The Maturation of Mechanical Properties of High Strength Concrete Exposed to Different
Moisture Conditions. In: Proceedings Third International Symposium on Utilization of High-Strength Concrete, June 20-24, Lillehammer, Norway, 1993, eds. Ivar Holland and Erik Sellevold, Vol 2, pp 1084-1091
[3] Weber, S. Reinhardt, H.W. Various curing methods applied to high- performance concrete with natural
and blended aggregates. In Proceedings of Fourth International Symposium on the Utilisation of High
Strength/High Performance concrete, Volume 3, Paris, May 29-3,1996, Editors F.de Larrard and R.Lacroix,
pp 1295-1303
[4] Guse,U. Dauerhaftigkeit von hochfestem Beton. DAfStb-Forschungs-kolloquium. Beitrge zum 32. Forschungskolloquium in Karlsruhe am 21.und 22.Mrz 1996, pp 99-106
[5] Weber, S. Nachbehandlungsunempfindlicher Hochleistungsbeton. Doctoral Dissertation, Stuttgart University, 1996
[6] DIN EN 196 Teil 1, Zemente (Ausgabe Mrz 1990)
[7] DIN 1048, Prfverfahren fr Beton, Teil 5: Festbeton, gesondert hergestellte Probekrper. Ausgabe Juni
1991
[8] Hennek, J. Liapor Franken, Private communication
[9] Chelouah, A. Private communication
[10] Reinhardt, H.W., Gaber, K.: Equivalent pore size characterising the pore size distribution of cement
mortar. Ceramic Transactions, Vol. 16, ed.S.Mindess, Am.Ceram.Soc., Westerville 1991, pp 319-335